Check registry
The 21 checks and controls
These are 21 proposed checks and controls, not 21 independent proofs. Several share a physiological cause and fail together; they are treated as correlated. No person is expected to complete every one.
See them in the synthetic walkthrough
Every entry is proposed. None is connected to qualified hardware, and none has been evaluated for accuracy. The second label shows only what the browser demonstration does with synthetic data.
| Key | Check | Kind | What it measures | Limit | Status |
|---|---|---|---|---|---|
1. capture | Trusted capture | Security control | Binds evidence to authenticated devices, fresh session challenges and timestamps. | A security control, not a biometric. Compromised capture software still needs separate defences. | Proposed Simulated in demo |
2. challenge | Randomised challenge response | Security control | An unpredictable, accessible prompt answered during capture. | Supports freshness only. A correct response alone is not proof of identity. | Proposed Simulated in demo |
3. depth | Facial depth | Sensor | Three-dimensional surface structure from a calibrated depth camera. | Masks, device errors and injected depth data require testing. | Proposed Unavailable · no hardware |
4. illumination | Active illumination | Sensor | Expected surface responses under approved, synchronised lighting. | Ordinary camera footage is not equivalent to a calibrated illumination system. | Proposed Unavailable · no hardware |
5. ppg | Contact optical pulse (PPG) | Physiological sensor | Pulsatile blood-volume changes from a validated contact sensor. | Not imaging of blood flow through the body. Motion and placement affect the signal. | Proposed Simulated in demo |
6. ecg | Electrical heartbeat (ECG) | Physiological sensor | Cardiac electrical activity from professionally designed electrode equipment. | Identity-related waveform features are research candidates only. | Proposed Simulated in demo |
7. correlation | ECG and PPG consistency | Cross-check | Whether electrical events and peripheral pulses have plausible timing. | Both share one physiological cause, so they are not independent guarantees. | Proposed Simulated in demo |
8. rppg | Camera pulse (rPPG) | Supporting evidence | Pulse-related skin colour changes across video frames. | Deepfakes can carry plausible pulse signals. A still image cannot measure a pulse. | Proposed Unavailable · no hardware |
9. vein | Vein pattern | Biometric sensor | A finger or palm vascular pattern from a purpose-built near-infrared device. | An anatomical pattern, distinct from a pulse waveform. Needs presentation-attack testing. | Proposed Not implemented |
10. iris | Iris pattern | Biometric sensor | Iris texture from suitable capture equipment and a tested matcher. | Not the same measurement as pupil movement or eye colour. | Proposed Unavailable · no hardware |
11. thermal | Thermal evidence | Sensor | Surface temperature patterns. | Influenced by surroundings; not proof of identity by itself. | Proposed Unavailable · no hardware |
12. identity | Identity evidence review | Workflow control | Establishes the claimed identity through an approved enrolment process. | Binds a person to a record; it does not detect a body. | Proposed Simulated in demo |
13. credential | Cryptographic credential | Authenticator | Control of a passkey or security key bound to the enrolled account. | Proves control of a credential. A non-biometric alternative is always required. | Proposed Simulated in demo |
14. intent | Intent and transaction approval | Workflow control | What the person is approving, bound to this session and action. | Presence does not imply consent; a match alone approves nothing. | Proposed Person-led step |
15. attendant | Trained attendant review | Human review | A documented way to resolve inconclusive results and offer the alternative route. | An operational control, not another biometric. | Proposed Person-led step |
16. document | ID document examination | Document check | Supported security features on an identity document, where authorised. | OCR only reads text; it does not prove authenticity, validity or ownership. | Proposed Not implemented |
17. face_match | Face to reference matching | Biometric comparison | One-to-one similarity between a fresh capture and a consented reference portrait. | Liveness and trustworthy capture remain separate tests. | Proposed Not implemented |
18. fingerprint | Fingerprint matching | Biometric comparison | A supported sensor with quality checks, matched against an authorised reference. | An absent reference or failed match means unresolved identity, never criminality. | Proposed Not implemented |
19. ten_print | Optional ten-print research | Research | Additional prints under a specific research protocol. | Overlaps with fingerprint matching; not ten independent guarantees. | Proposed Not implemented |
20. respiration | Respiratory motion | Sensor | External breathing-related movement seen by a camera or depth sensor. | A webcam cannot see through a rib cage; recorded motion may be replayed. | Proposed Unavailable · no hardware |
21. flow | Direct blood-flow research | Research sensor | Blood movement measured with specialist Doppler or other validated flow equipment. | Needs its own hardware and evaluation; not a software upgrade to a webcam. | Proposed Unavailable · no hardware |
Out of scope, always
Blood typing, platelets, lie detection, public-camera identification and any global biometric tracking index are excluded from this research. A physiological reaction is not a lie detector, and a failed reading is never evidence of wrongdoing.
Next step: try the walkthrough or return to the overview.